logicbeat/ qwen3.8-27B_GSQ_RCO:latest

3,006 1 month ago

Dynamically quantized only 12GB MTP model with precision close to the base model picked from hugging face ISTA-DASLab/Qwen3.8-27B-GSQ-RCO-IQ3_S-GGUF, highly efficient model

vision
... /
params
83bb9f945317 ยท 98B
{
"draft_num_predict": 3,
"min_p": 0,
"presence_penalty": 0,
"repeat_penalty": 1,
"top_k": 20,
"top_p": 0.95
}